The objectives of this study were to (1) estimate the clinical status of the mammary glands and (2) compare it with the bacteriological findings, the total cell content (TCC) and its percentage of polymorphonuclear leucocytes (PMNLs) and pH in colostrum and milk secretion of sows on 2 different feeding regimes, high versus low, during late pregnancy. The milk samples were collected from both agalactia post partum (APP) sows and clinically healthy sows. Sows with a rectal temperature exceeding 39.5°C within 48 h after parturition were considered to be diseased in APP and treated medically. The sows were sampled on days 1, 3, 8 and 22 of lactation during 6 consecutive lactations. Irrespective of feeding regimes, 49 out of 77 lactations among the APP sows and 15 out of 96 lactations among the clinically healthy sows revealed E. coli in pure cultures with a concomitant TCC exceeding 10×106 cell/ml already on the first day of lactation. The healthy sows with E. coli infection were denominated as being subclinically infected sows. The intensity in growth of E. coli successively declined, and the bacteria were finally eliminated between days 3 and 8 of lactation. The TCC were 82×106 cells/ml and 157×l06 cells/ml in the clinically and subclinically E. coli infected glands, respectively, on the first day of sampling. The TCC declined gradually in both groups of sows, but was still higher than in bacteriologically negative milk on day 22 of lactation. The percentages of PMNLs were 66% and 79% in clinically and subclinically infected glands, respectively, on day 1 of lactation, thereafter decreasing to approximately 50% on day 22 of lactation in both groups of sows. In APP sows, swelling, reddening and/or soreness were registered in 38 out of 87 mammary glands with E. coli mastitis on the first sampling occasion. The TCC in bacteriologically negative colostrum and milk collected from APP sows on day 1 of lactation was significantly higher, 2.27×106 cells/ml, when compared with the TCC in bacteriologically negative milk secretion from the clinically healthy or subclinically infected sows, 1.38×106 cells/ml versus 1.51×106 cells/ml, respectively. The PMNLs were higher on day 1 in clinically healthy sows, 59.6%, than in subclinically infected and APP sows (43.5% and 48.3% respectively). The pH in secretion from clinically or subclinically E. coli infected glands (6.57 versus 6.46) were higher than in bacteriologically negative colostrum samples (6.29) from clinically diseased sows on the first day of sampling. On day 22 of lactation, pH-values had stabilized on a level of approximately 7.00 in all milk samples from earlier bacteriologically positive or negative mammary glands. The 2 feeding regimes, low versus high, were not found to influence TCC, PMNLs or pH except for TCC in bacteriologically negative samples of APP sows (2.69 versus 3.62). The lactation number influenced the PMNLs in both groups of sows with E. coli infected mammary glands, and both the TCC and PMNLs in bacteriologically negative colostrum and milk.